If you already have a dataset you are familiar with and have cleaned, you can use it for your assignments
Only two rules: (1) Must have > 500 observations, and (2) must already have it cleaned (i.e., variables have clear names, variables are labeled, values are labeled, you know how much missing data there is)
If not, I provide some semi-cleaned datasets you can use for your assignments
Look at the descriptive statistics tables of these datasets in the Datasets for Assignments folder on Brightspace (under 00-Misc)
General Social Survey 1972 – 2021 (GSS)
General Social Survey (GSS) 1972 to 2021
Cross-sectional survey fielded every two years in the U.S.
Mainly tracks social and political attitudes and opinions
Special modules in some years with questions about specific topics (e.g., environmental attitudes, mental health, etc)
6,108 variables in full dataset; 193 semi-cleaned in the course dataset
My advice: keep only two countries of interest for class assignments
Use the country indicator as your binary IV for assignments
Creating scales for continuous variables
It can be challenging to find continuous variables in health and social science datasets (that is one reason you need this class!)
Scales and indexes are some of the most common forms of continuous variables
Find a few binary or ordinal variables that you think are measuring a single underlying construct and propose in Assignment 2 to combine them into a scale or index
I will help guide you as to whether this will work or not
There are lots of sophisticated ways to create scales; for the purpose of this class, you can just add the variables* together into a “summated scale”
In the course datasets I provide, many scales have already been created for you. These variable names usually end in SS (“summated scale”)
*If all items are measured in the same way, often useful to divide by total # of items to retain interpretation
Interpreting scales
A difficulty with scales is they often have no natural metric so can be awkward to interpret
I suggest interpreting the mean value as “average” on that scale, ± 1 SD “somewhat” more or less on that construct, and ± 2 SDs as “very” indicative of that construct (or lack of it)
It can be very useful to label these values on graphs, to use them as places to make predictions, to aid your interpretations, etc.
E.g., For a depression scale:
Depression Scale
-2 SDs
-1 SD
Mean
+1 SD
+2 SDs
“very low depressive symptoms”
“somewhat less depressed than average”
“average”
“somewhat more depressed than average”
“very high depressive symptoms”
Recoding variables into a natural metric
Another strategy for making continuous variables is to recode ordinal variables where the response options have a natural metric (but were not coded that way)
E.g., Recode religious attendance variable into days per month (e.g., assign “2-3 times per month” a value of 2.5. Assign “every week” a value of 4. Etc.)
(cda_gss.dta | GSS 1972-2021 CDA - Categorical Data Analysis | date created 202
> 3)
fre attend
attend -- how often r attends religious services
------------------------------------------------------------------------------
| Freq. Percent Valid Cum.
---------------------------------+--------------------------------------------
Valid 0 never | 12706 18.46 18.64 18.64
1 less than once a year | 5409 7.86 7.93 26.57
2 about once or twice a | 8951 13.00 13.13 39.70
year |
3 several times a year | 8406 12.21 12.33 52.03
4 about once a month | 4674 6.79 6.86 58.88
5 2-3 times a month | 5913 8.59 8.67 67.55
6 nearly every week | 3842 5.58 5.64 73.19
7 every week | 13218 19.20 19.39 92.58
8 several times a week | 5062 7.35 7.42 100.00
Total | 68181 99.03 100.00
Missing .d | 595 0.86
.i | 40 0.06
.n | 3 0.00
.s | 27 0.04
Total | 665 0.97
Total | 68846 100.00
------------------------------------------------------------------------------
library(haven) # import Stata data and labelslibrary(sjmisc) # frequency tablesgss <-read_dta("https://tdmize.github.io/data/data/cda_gss.dta")frq(gss, attend)
how often r attends religious services (attend) <numeric>
# total N=68846 valid N=68181 mean=3.71 sd=2.74
Value | Label | N | Raw % | Valid % | Cum. %
---------------------------------------------------------------------
0 | never | 12706 | 18.46 | 18.64 | 18.64
1 | less than once a year | 5409 | 7.86 | 7.93 | 26.57
2 | about once or twice a year | 8951 | 13.00 | 13.13 | 39.70
3 | several times a year | 8406 | 12.21 | 12.33 | 52.03
4 | about once a month | 4674 | 6.79 | 6.86 | 58.88
5 | 2-3 times a month | 5913 | 8.59 | 8.67 | 67.55
6 | nearly every week | 3842 | 5.58 | 5.64 | 73.19
7 | every week | 13218 | 19.20 | 19.39 | 92.58
8 | several times a week | 5062 | 7.35 | 7.42 | 100.00
<NA> | <NA> | 665 | 0.97 | <NA> | <NA>
